SEEDS Internship : Megacity Megaquakes: advancing seismic monitoring in urban environments

Traditionally, seismic monitoring stations avoid urban environments because anthropogenic noise contaminates seismic records. However, to better understand seismic hazard and risk in urban areas, it is vital to have seismic sensors in urban areas to map seismogenic faults and to probe shallow geological structures that can affect ground motions. This issue is exemplified by many megacities around the world lying on hazardous plate boundary faults (e.g., Los Angeles, U.S.; Santiago, Chile; Tokyo, Japan; Istanbul, Türkiye). Recent advances in artificial intelligence and machine learning have fortunately made it feasible to remove anthropogenic signals from seismic data to reveal natural earthquake signals. Such “denoising” models already exist, and this project will test their success. In this project, the intern will take advantage of a brand-new pool of broadband seismometers at UCL by deploying these instruments around the campus in central London, a particularly seismically noisy environment. You will download and quality-control data recorded by these sensors every few days, and will test existing denoising neural network models. You will directly interact with a diverse group of seismologists (including PhD students) at UCL and Birkbeck. Your work will play a key role in improving seismic monitoring in urban environments – an underexplored frontier in earth science and seismology. You will develop skills in machine learning, analysing big datasets, and signal processing.

Work plan: Full-time/part-time

- Training in using the seismic hardware and seismic data analysis using Python

- Deployment of the seismometers around UCL

- Downloading, quality-control of data, and applying a deep denoising workflow.

- Recovery of seismometers; project wrap-up and debrief